JP2868864B2 - Inspection equipment for printed wiring boards - Google Patents

Description

Description: BACKGROUND OF THE INVENTION The present invention relates to an inspection apparatus for detecting a defect shape of a wiring pattern on a printed wiring board, and more particularly, to an inspection apparatus for preventing a false report from around a through hole. About.

[Conventional technology]

In the manufacturing process of a wiring pattern on a printed wiring board, a defect shape such as disconnection, chipping (line thinning), short circuit, defective line-to-line connection, or rubber adhesion may occur. Therefore, various inspection devices for detecting such a defect shape have been proposed.

On the other hand, as shown in FIG. 13, the printed wiring board 10 has a through hole 90 and a wiring pattern 9. The wiring pattern 9 includes a land 91 formed in an annular shape at the periphery of the opening of the through hole 90 and a line pattern 92 connecting between circuit terminals (not shown). The wiring pattern 9 is formed by a metal plating layer.

And, in general, the width of the land 91 (for example, about 50 μm)
Are formed narrower than the width of the line pattern 92 (for example, about 150 μm). Therefore, as shown in FIG. 14, when the land 91 is formed slightly off the axial center of the through hole 90, a narrow portion 911 and a wide portion 912 are formed. Further, the narrow portion 911 also occurs due to a small seat width due to poor etching. Therefore, when the above-mentioned land is inspected by the inspection device,
As shown in FIG. 15, the inspection data 94 is detected. In the inspection data 94, the narrow portion 911 of the land 91 is considerably narrower than the width of the line pattern 92, and the line pattern 92 is recognized as having a defect shape. Therefore, the narrow portion 911 is detected as being in the missing (line thinning) state 941.

However, the narrow portion 911 is not in a disconnected state as shown in FIG. 14, and there is no practical problem. As described above, false information is likely to occur in the inspection data obtained by the conventional inspection device. Such false information is particularly large for the land 91 formed around the through hole.

To cope with the above problem, there is an inspection apparatus in which mask processing is performed on a through-hole portion, and inspection around the through-hole is not performed when inspecting a line pattern, so that erroneous determination around the through-hole is eliminated.

The inspection apparatus includes a CCD camera 11 that reads a wiring pattern 101 on a printed wiring board 10 as shown in FIG.
A / D converter 12 for converting an analog signal of CCD camera 11 into a digital image signal, mask section 81, defect shape recognition section 851, and signals from mask section 81 and defect shape recognition section are collated. It comprises a through-hole mask collating unit 841, a data compiling unit 15 for compiling the collation results, and a terminal 16 as display means.

The mask unit 81 includes a through-hole mask creating unit 811 that forms a through-hole mask based on a binary image signal from the A / D converter 12 based on a standard wiring pattern of a printed wiring board. The memory unit 812 stores a through-hole mask, and a reading unit 813 reads out the through-hole mask at the time of inspection.

The through-hole mask was created by the through-hole mask creating unit before the wiring pattern was formed. It is created by taking an image around the through hole using a printed circuit board that does not cause breakage on the binary image in the case.

In this way, the positional relationship between the hole of the through hole and the land is recognized, and a mask having a sufficient size is created so that no false alarm is output to all the detection systems during an actual inspection. The detection system is a system (corresponding to the defect shape recognition units 851, 852, and 853 in FIG. 16) that performs detection according to each defect shape such as disconnection, chipping, and short-circuit in a line pattern.

Further, the reading unit 813 corrects the defect shape recognition positions of all the detection systems at the time of reading so that a through-hole mask is applied moderately.

Next, the defect shape recognition unit 851 processes the image signal from the A / D converter 12 with a spatial filter, and recognizes (detects) the defect shape related to the disconnection by the feature extraction method.

In the same figure, the other defect shape recognition units 852 and 853
Similarly, the image signal is processed by a spatial filter to recognize a defect shape related to a missing or short circuit.

Each of the spatial filters is designed to pass only an image signal corresponding to a defect shape such as a disconnection or a short-circuit. In addition, this defect shape recognition unit uses A / D
If the binary image signal from the converter 12 is data (defect candidate data) temporarily determined to be a defect shape, the defect candidate data is transmitted to the through-hole mask collation unit.

Further, the through-hole mask collation unit 841 collates the defect candidate data from the defect shape recognition unit 851 with the through-hole mask from the read unit 813 of the mask unit 81. At this time, the reading of the through-hole mask is corrected so that the through-hole mask is applied to the middle of the defect candidate data relating to the disconnection of the defect shape recognition unit 851, that is, the center of the matrix of the spatial filter for defect shape recognition. I do it.

Then, as a result of the above-mentioned collation in the through-hole mask collation unit 841, when it is determined that the defect candidate data has a true defect shape, a signal to that effect is sent to the data totaling unit 15 and the terminal 16. That is, if the defect candidate data is not inside the through-hole mask, it is recognized as a true defect shape.

These operations are also performed between the defect shape recognition units 852 and 853 and the through-hole / mask matching units 842 and 843, and defect shapes such as chipping and short-circuit are inspected.

[Problem to be solved]

However, the detection of defect candidate data in the defect shape recognition unit is a method of extracting a feature using a spatial filter, and thus a rule for extracting a feature is set in a window of the spatial filter. For this reason, it is not always possible to set a rule so that a defective portion always fits the center of the window.

On the other hand, since the through-hole mask is created around the hole, when the spatial filter rule is applied around the through-hole on the defect shape recognition spatial filter, the mask is placed at the position corresponding to the hole. It can be said that applying the mask is an accurate method of applying the mask.

Figure 17 shows a hole in the window 96 of the spatial filter.
960 shows a state where it was located. This figure shows a chipped state. That is, in this case, there are an image signal of the pattern image 962 of the wiring pattern and an image signal of the base material image 961 of the printed wiring board, and the hole 960 which is assumed to be a hole is located. In any position. That is, the hole portion 960 is not at the center position.

On the other hand, FIG. 18 shows a pattern image 972 and a base material image 971,
The position assumed to be the hole is the point indicated by 970. The figure shows the case of the disconnection state.

These figures show that when the rule is set for windows of the same size, the positions of the hole portions are arbitrary.

As described above, since the positions of the holes in the window are different, even if a uniform through-hole mask is transmitted from the mask readout unit 813 to the through-hole mask collation unit 841, the collation accuracy is high. bad. In other words, although the defect candidate data is not a true defect shape,
A false report is issued as a true defect shape in the through-hole mask collation unit.

Therefore, in order to prevent the false alarm, it is conceivable to send data of a large through-hole mask to a through-hole mask collation unit. However, in this case, conversely,
The true defect shape generated in the line pattern near the through hole will be missed.

In view of such a conventional problem, the present invention proposes a through hole with an optimal position and size for each defect candidate data.
An object of the present invention is to provide a printed wiring board inspection apparatus capable of collating a mask and preventing a false report from around a through hole.

[Solutions to solve the problem]

The present invention relates to an imaging device that images a wiring pattern on a printed wiring board, a mask unit that creates and stores a through-hole mask from an image signal of the imaging device, and compares the image signal with the through-hole mask. And a data counting unit that counts the defect data of the defect detection unit. The mask unit extracts a hole portion of the through hole of the printed wiring board and arbitrarily enlarges or reduces the extracted hole image. The through hole basic mask creating unit, and the extraction unit created by the through hole basic mask creating unit. And a mask memory unit for storing the hole image as a through-hole mask image. Further, the defect detection unit recognizes the defect shape from the image signal using a spatial filter in which a rule assuming the defect shape is set,
A defect shape recognition unit that outputs the defect candidate data, and a matching position between the basic mask image output from the mask memory unit at the time of inspection and the defect candidate data of the defect shape recognition unit are determined with respect to an input image unique to each defect shape recognition unit. A through-hole mask correction unit that moves the through-hole mask side to match the shape recognition position and expands the size of the adaptation area as necessary, and a correction output from the through-hole mask correction unit And a through-hole mask comparison unit that compares the completed through-hole mask with the defect candidate data output from the defect shape recognition unit. Further, the defect detection units are arranged according to the number of types of defect shapes.

The most remarkable feature of the present invention is that a mask memory unit having the through-hole basic mask creating unit and a defect detecting unit having the through-hole mask correcting unit are provided.

In the basic mask creating section, only the through hole portion of the printed wiring board is extracted, and the extracted hole image is arbitrarily enlarged or reduced. The extracted hole image is created, for example, by filtering an image of a drill board captured by the imaging device and extracting only the hole portion. The drill board is a plate in which a hole for a through hole is formed in a substrate for a printed wiring board.

In addition, the extracted hole image is reduced or enlarged to an arbitrary size in advance as necessary so that the extracted hole image can be expanded in the through hole mask correction unit to an appropriate size for the defect candidate data of the defect shape recognition unit. Enlarge it. Normally, the dimension is set to the minimum necessary size, expanded as needed by the through-hole mask correction unit, and sent to the through-hole mask comparison unit.

As described above, the extracted hole image arbitrarily sized is stored in the through-hole mask memory unit and transmitted to the through-hole mask correction unit when collating with the defect candidate data of the defect shape recognition unit. I do.

The defect detection unit is provided for each defect detection system corresponding to a defect shape such as disconnection, chipping, or short-circuit. Therefore, the defect detection units are provided by the number of the defect shapes to be detected. Each defect detection unit has a defect shape recognition unit, a through-hole mask correction unit, and a through-hole mask collation unit according to the defect shape to be detected. That is, the defect detection unit includes a set of defect shape recognition units, a through-hole mask correction unit, and a through-hole mask comparison unit for one type of defect shape.

The defect shape recognition unit processes the binary image signal sent from the A / D converter by using a spatial filter as in the conventional case. That is, using a spatial filter designed according to the type of the defect shape, defect candidate data such as disconnection or chipping is recognized. Then, the defect candidate data is sent to a through-hole mask collation unit.

On the other hand, in the through-hole mask correction unit, in order to adapt the through-hole mask of the corresponding defect shape to the defect candidate data detected by the defect shape recognition unit of the same group (for example, for disconnection detection), The basic mask image is called out from the through-hole mask memory section.

What is important here is that the basic mask image called from the through-hole mask memory unit performs shape recognition on the input image unique to each defect shape recognition unit before collation with the defect candidate data of the defect shape recognition unit. Moving the through-hole / mask side according to the position. Also,
The purpose is to expand the size of the matching area according to the defect shape of the defect candidate data.

As described above, examples of means for moving the matching position according to the defect candidate data and means for expanding the size of the matching area include the following.

That is, as such means, the same size as the spatial filter for recognizing the defect shape (the resolution may not be particularly the same.
Pixel matrix spatial filter. For the mask correction, a spatial filter having a resolution of 20 μm and a spatial filter of a 10 × 10 pixel matrix is used. In the spatial filter of this through-hole mask correction unit, the outputs of the points registered on the matrix are all connected by an OR condition, and the through-hole mask matching unit is arbitrarily registered on the spatial filter matrix. The above functions can be realized. For example, ABC, DEF,
If a basic through-hole mask of positive logic is input to any of BDEFH for the nine matrices of GHI, a corrected mask is obtained with negative logic.

The through-hole mask comparison unit collates the defect candidate data from the defect shape recognition unit with the corrected through-hole mask from the through-hole mask correction unit. When the defect candidate data exists in the through-hole mask, it is determined that the defect shape is not a true defect shape. In the opposite case, it is transmitted to the data totaling unit as a true defect shape.

In the present invention, the defect shape occurring in the vicinity of the through hole means a defect shape in a line pattern near the through hole other than the land provided on the periphery of the opening of the through hole. Further, "around the through hole" means a run pattern near the land and the through hole.

[Action and effect]

In the present invention, since the through-hole basic mask creating section is provided in the mask section, the extracted hole image can be arbitrarily sized, and the basic size of the through-hole mask to be transmitted to the through-hole mask correction section can be reduced. It can be arbitrarily selected. For this reason, the basic size can be set finely to the minimum dimension and the like, and the fact that the through hole mask is large as in the prior art cannot detect the true defect shape generated near the through hole. Absent.

In addition, since the extracted hole image is processed like a bitmap,
Even if the through-hole has an irregular shape, the recognition processing of the through-hole can be easily performed.

In addition, the through-hole mask correction unit can shift the conforming position in accordance with the defect detection system such as a disconnection and expand the size of the conforming area as necessary. Accordingly, the degree of engagement of the through-hole mask can be precisely and finely adjusted.

Therefore, according to the present invention, it is possible to collate a through hole mask having an optimum position and size with respect to each defect candidate data, and to prevent a false report from around the through hole. An inspection device can be provided.

〔Example〕

First Embodiment A printed wiring board inspection apparatus according to an embodiment of the present invention will be described with reference to FIGS. 1 to 8C.

That is, as shown in FIG. 1 in its entirety, the inspection apparatus of the present embodiment includes a CCD camera 11 as an imaging device for imaging a wiring pattern 101 on a printed wiring board 10, an A / D converter 12, and a CCD camera.
A mask unit 2 for creating and storing a through-hole mask from the binary image signal of the CD camera 11, a defect detecting unit 3 for comparing the image signal with the through-hole mask, and a defect of the defect detecting unit 3 Data tabulation unit 15 for tabulating data
And terminal 16.

The mask unit 2 includes a through-hole basic mask creating unit 21 that extracts only a hole portion of a through-hole of the printed wiring board 10 and arbitrarily enlarges or reduces the extracted hole image. A through-hole mask memory unit 22 stores the created extracted hole image as a through-hole mask image.

The defect detection unit 3 includes a defect shape recognition unit 31, a through-hole mask correction unit 35, and a through-hole mask comparison unit 36.

The defect detector 3 is used as a disconnection defect detection system. As shown in FIG. 1, the defect detection unit includes a defect detection unit 3B for a chip, a defect detection unit 3C for a short circuit, and a defect detection unit 3D for a defective line, as shown in FIG.

The defect shape recognition unit 31 determines the presence or absence of the defect shape from the binary image signal from the A / D converter 12 using a spatial filter in which a rule assuming the defect shape is set. If the answer is YES, defect candidate data is output.

The through-hole mask correction unit 35 determines the matching position between the basic mask image output from the through-hole mask memory unit 22 and the defect candidate data at the time of inspection with respect to an input image unique to each defect shape recognition unit. The matching is achieved by moving the through hole / mask side in accordance with the shape recognition position. At the same time, the size of the matching area of the through-hole mask is expanded as necessary.
In the through-hole mask correction unit 35, a through-hole mask relating to the disconnection is read from the through-hole mask memory unit 22.

Next, in the through-hole mask collation unit 36, the defect candidate data from the defect shape recognition unit 31 and the through-hole mask from the through-hole mask correction unit 35 are collated. Check if you are inside the hole mask. If the defect candidate data is included in the through-hole mask, it is determined that the defect shape is not a true defect shape.

On the other hand, in the case opposite to the above, it is determined that the shape is a true defect. Then, these data are transmitted to the data totaling unit 15 and the terminal 16.

The operation of the defect detection unit 3 is different from that of the other defect detection units 3.
In B, 3C and 3D, except that the defect recognition shape (chip, short, etc.) and the through-hole mask correction rule are different,
The same is true.

 Next, the details of each part will be described.

First, a through-hole basic mask creating section 21 of the mask section 2
In FIG. 2, a basic mask is created as shown in FIG. That is, the periphery of the through hole is imaged by the CCD camera 11 using a drill board having a hole for the through hole or a printed wiring board having no break in the land portion of the through hole. Then, the image data is converted into a binary image signal at an arbitrary threshold level by the A / D converter 12.
At this time, the image signal includes the through-hole image 4 and the unnecessary image 41 such as dirt and dust in the periphery, as shown in FIG. 2A.

Therefore, as shown in B, images smaller than the set shape are cut using a spatial filter 451 for removing unnecessary images 41 such as dirt and dust. Thus, an extraction hole image 40 shown in C is obtained.
A dotted circle 42 around the extracted hole image 40 is the magnitude of the through-hole image signal 4 indicated by A.

The extracted hole image 40 obtained as described above is restored to a restored image 401 having the same diameter as the original through-hole image 4 as shown at E by using a restoration spatial filter 452 as shown at D in FIG. it can. Further, this restored image 401 can be made into an enlarged image 403 as shown by G by using the spatial filter for enlargement 453 shown by F.

Spatial filter for restoration 452, Spatial filter for enlargement
453 is a rule that if there is even one pixel input in the specified area, an output is generated.
It can be called a connected circuit.

Also, the dirt removal spatial filter 451 has a rule that allows only inputs larger than the specified area to pass.
This is a circuit in which all the specified area inputs are AND-connected.

As described above, in the through-hole basic mask creating unit 21, the extracted hole image 40 is enlarged or reduced to an arbitrary size.
Of course, the size may be the same as the extracted hole image 40. Thereby, a basic mask image is created. The basic mask image is stored in the through-hole mask memory unit 22, and is read out by the through-hole mask correction unit 35 at the time of inspection.

Next, in the defect shape recognition unit 31, a binary image signal is input from the A / D converter 12 at the time of inspection, and the image signal is recognized as a broken line portion temporarily using a spatial filter designed for detecting a broken line. It is detected whether there is “defect candidate data”. When there is defect candidate data, the data is output to the through-hole mask collation unit 36.

FIG. 3 shows a specific configuration example of the spatial filter.
As is known from the figure, the spatial filter receives a binary image as input, and is constituted by a shift register in the horizontal direction and a line memory in the vertical direction. Then, the output from the shift register is inverted, and the defect candidate data is output through a filter of an AND condition. That is, in the defect shape recognition unit 31, the disconnection shape is implemented by hardware using this method. By using this spatial filter, an inspection for disconnection is performed.

Next, in the through-hole mask correction unit 35,
As shown in FIGS. 4A and 4B, the correction of the through-hole mask is performed.

That is, as shown in FIG. 4A, when there is a pseudo disconnection portion 511 in the through-hole land 51, the image signal is transmitted to the substrate image 531 and the pattern image 532 and the defective portion within the window 53 of the disconnection shape recognition spatial filter. Compatible with 533.

Therefore, in the through-hole mask correction unit,
As shown in Fig. 4B, the through-hole mask memory section 22
Then, the basic mask 44 is called and input.

In the correction spatial filter window 460, data taking into account the design of the spatial filter 53 is set. That is, when the pseudo disconnection around the through hole is matched in the window 53, it is checked where the hole is located, and the corresponding portion is registered in the window 460. This registration area 46 is illustrated in FIG. 4B.

By adjusting the position and size of the registration area 46, the basic mask 44 can be optimally applied to the defect candidate data.

If at least one point of the basic mask matches the registration area 46 in the correction spatial filter window 460, an output is generated. This output is the corrected through-hole mask
The data is sent to the through-hole mask collation unit 36.

FIG. 5 shows a state where the outputs from the defect shape recognition unit 31 and the through-hole mask correction unit 35 are collated by the through-hole mask collation unit 36 using an AND element.

In the figure, a base material image 61 and a pattern image 62 are set in a window 6 of the defect detection spatial filter of the defect shape recognition unit. The hole position when the rule of this spatial filter conforms around the through hole is the hole position.
Indicated by On the other hand, the through-hole mask correction unit 35
In the spatial filter for correcting the mask matching position of
The registration area 46, which is a matching point of the basic through-hole mask, is located at a certain position in the window 460. The registration area 46 is located in the hole
This is a position corresponding to 63 points.

Therefore, the outputs from the defect shape recognition unit 31 and the through-hole mask correction unit 35 enter the through-hole mask comparison unit 36 as shown in FIG. And the through hole
When the basic mask matches the position of the hole portion 63, that is, the position of the registration area 46 of the through-hole mask, the mask collating unit 36 outputs 0 (zero). That is, it is determined that the defect candidate data of the defect shape recognition unit 31 is not a true defect shape. If there is no match, 1 is output and it is determined that the shape is a true defect.

On the other hand, to prevent false alarms from around the through-hole,
It is necessary to enlarge the through-hole mask, but it is necessary to keep the through-hole mask small in order to detect defects around the through-hole. The size of the through-hole mask has a different optimal size depending on the type of the defect shape. Therefore, in this example, the size of the through-hole mask is changed according to the type of the defect shape.

That is, as shown in FIG. 6, processing is performed by applying the spatial filter for correcting the mask matching position. That is,
The through-hole mask is expanded around the position of the hole portion 63 in the window 460, and the matching area 461 is set large. As a result, the through-hole mask can be expanded substantially with respect to the input through-hole mask.

As is known from the above description, in this example, the input through-hole mask is set to the minimum diameter in the through-hole basic mask creating unit 21, and thereafter, the through-hole mask is enlarged according to the defect shape. In this way, by setting a large matching area of the through-hole mask with the defect matching point as the center, the through-hole mask is set.
The mask correction unit can substantially enlarge the through-hole mask with respect to the input through-hole mask.

In FIG. 6, if the through-hole mask matches somewhere in the designated area of the spatial filter for mask correction, the defect detection rate is output as zero (0) as not being a true defect shape, and the other is 1 Is output.

The spatial filter of the defect shape recognition unit uses, for example, a resolution of 10 μm and a window size of 20 × 20 pixels (400 cells). The spatial filter for the through-hole mask in the through-hole mask correction unit uses, for example, a resolution of 20 μm and a window size of 10 × 10 pixels (100 cells).

Next, FIGS. 7 to 8C show processing when defect candidate data relating to disconnection is input to the defect shape recognition unit 31. FIG.

FIG. 7 shows a wide portion of the through hole land image 7.
When there are 71 and narrow portion 72, base material image in narrow portion 72
Based on the pattern image 61 and the pattern image 62, defect candidate data is output assuming that there is a disconnection shape 64. On the other hand, below the through-hole image, there is a true disconnection defect between the wiring portion 73 and the land 71, which is output as a defect portion 68 between the base material image 65 and the pattern image 66.

FIG. 8A shows a design of a spatial filter for recognizing the disconnection shape at this time.

On the other hand, a spatial filter for through-hole mask correction is set as shown in FIG. 8B.

The basic through-holes are shown in the shaded pixels in Fig. 8B.
If the mask matches, the mask processing is performed on the defect candidate data.

Therefore, in the above case, the defect candidate data can be sufficiently masked with a mask that covers the through hole with respect to the pseudo disconnection shape 64 in the through hole, and the disconnection shape 64 must be a true defect shape. I can judge.

Also, at the same time, the true defect part under the through hole land
Since 68 is not masked by the through-hole mask of the through-hole mask correction unit, it can be determined that a disconnection has occurred in the line pattern.

In the above, in the through-hole mask correction unit, as shown in FIG.
When a mask correction spatial filter is used,
Compared to the case of Fig. 8B, a through-hole mask with a size of 3 pixels or more per radius of the through-hole is required.
In addition, when such a large through-hole mask is used at all times, the true defective portion 68 is masked, and the defect cannot be detected.

As is known from the above, according to this example, the matching position between the basic mask image and the defect candidate data of the defect shape recognition unit is determined by the through-hole mask correction unit 35.
By moving the mask side, the matching is achieved, and the size of the matching area is expanded as necessary. The through-hole basic mask creating unit is configured so that the extracted hole image can be arbitrarily enlarged or reduced.

Therefore, unlike the related art, there is no possibility that a true defect shape cannot be detected due to the large through-hole mask. Therefore, a through-hole mask having an optimum position and size can be individually checked with respect to each defect candidate data, and a false report from around the through-hole can be prevented.

Second Embodiment This embodiment shows a case where the defect shape is a chipped (thin line) state, which will be described with reference to FIGS. 9A to 11C.

That is, as shown in FIG.
If a part of the line has a thinned portion 515, the base material image appears in the window 53 of the spatial filter for defect shape determination.
Based on the pattern image 61 and the pattern image 62, the defect candidate data is output on the assumption that the line thin shape 64 exists.

Then, as shown in FIG. 9B, the basic mask 44 is input to the spatial filter 460 of the through-hole mask correction unit. In the window of the spatial filter, when the spatial filter 53 for defect shape determination conforms to the thin line portion 515, data registration is performed at a point having the same positional relationship as the base material portion 61 where the hole is predicted to be located. Need to be done. This is shown in FIG. 9B.

In FIG. 10, when there is a thin line portion in the through-hole land image 7, the defect candidate data is output assuming that there is a thin line portion 64 by the base image 61 and the pattern image 62. Then, as shown in the figure, a basic through-hole mask 751 indicated by a stepped dotted line is applied through a mask correction spatial filter of a through-hole mask correction unit.

In the above case, the thin line portion in the through hole is recognized as a defect candidate by the spatial filter of the defect shape recognition unit of the design shown in FIG. 11A. On the other hand, the basic through-hole mask image is processed by the through-hole mask correction spatial filter shown in FIG. 11B in the through-hole mask correction unit.
That is, if the basic mask is somewhere in the designated portion, the defect candidate data is masked, and it is determined that the defect is not a true defect shape.

In the above description, if a spatial filter for through-hole mask correction is used in the through-hole mask correction unit as shown in FIG. Will be applied. Therefore, according to this rule, the thin defect candidate data cannot be masked using the above-described through-hole mask, and a false report appears. for that reason,
Further, it is necessary to make the through-hole mask larger, and if it is made larger, the above problem occurs.

As described above, in this embodiment, the same effects as in the first embodiment can be obtained.

[Brief description of the drawings]

1 to 8C show the inspection apparatus of the first embodiment, FIG. 1 is a block diagram showing the whole of the inspection apparatus, and FIG. 2 shows extraction and enlargement of an extracted hole image in a through-hole basic mask creating unit. Fig. 3, Fig. 3 is an explanatory view of the spatial filter of the defect shape recognition unit, Fig. 4A and Fig. 4B are images showing disconnection images, Fig. 5 is positioning adaptation in the defect shape recognition unit and through-hole mask correction unit , FIG. 6 is an explanatory diagram of a conforming area in the defect shape recognition unit and the through-hole mask correction unit, FIG. 7 is an explanatory diagram of a through-hole line pattern image relating to disconnection, and FIGS. 8A to 8C are spaces. FIG. 9A to FIG. 11C show the application of the filter, and FIG.
FIG. 9 and FIG. 9B are explanatory diagrams of the image relating to the chipping, FIG.
Fig. A to Fig. 11C are explanatory diagrams of applying a spatial filter, Fig. 12 is an explanatory diagram of a spatial filter of a through-hole mask correction unit,
13 to 18 show a conventional example, FIG. 13 is a plan view of a through hole and a line pattern in a printed wiring board, FIG. 14 is a plan view of a through hole land portion in a thin line state, and FIG. FIG. 13 is an image signal of a through-hole land portion, FIG. 16 is a block diagram of an inspection device, and FIGS.
The figure is an image of defect candidate data. 4 ... through-hole image, 40 ... extracted hole image, 44 ... basic mask, 50, 53 ... window, 61 ... substrate image, 62 ... pattern image, 63 ... hole part, 7 ... through Holland statue,

Claims (1)

(57) [Claims] An imaging device for imaging a wiring pattern on a printed wiring board, a mask unit for creating and storing a through-hole mask from an image signal of the imaging device, and an image signal and the through-hole mask The mask unit extracts only a hole portion of a through hole of a printed wiring board and arbitrarily extracts the extracted hole image. Enlargement,
The through hole basic mask creating unit to be reduced and the extracted hole image created by the through hole basic mask creating unit
A mask memory unit for storing the mask image as a mask image; and the defect detection unit recognizes the defect shape from the image signal by using a spatial filter in which a rule assuming the defect shape is set. A defect shape recognition unit that outputs the defect candidate data, and a matching position between the basic mask image output from the mask memory unit during inspection and the defect candidate data of the defect shape recognition unit are determined with respect to an input image unique to each defect shape recognition unit. A through-hole mask correction unit that matches by moving the through-hole mask side according to the shape recognition position and expands the size of the adaptation area as necessary, and a correction output from the through-hole mask correction unit And a through-hole mask matching unit that matches the used through-hole mask with the defect candidate data output from the defect shape recognition unit. In addition, the inspection apparatus of printed wiring board, characterized in that the defect detecting section are disposed in accordance with the number of kinds of defects shape.